Facial Expression Recognition Using Thermal Image Processing and Efficient Preparation of Training-data

Abstract

Using our previously developed system, we investigated the influence of training data on the facial expression accuracy using the training data of "taro" for the intentional facial expressions of "angry," "sad," and "surprised," and the training data of respective pronunciation for the intentional facial expressions of "happy" and "neutral." Using the proposed method, the facial expressions were discriminable with average accuracy of 72.4% for "taro," "koji" and "tsubasa", for the three facial expressions of "happy," "neutral," and "other".

Facial Expressions Recognition in Thermal Images based on Deep Learning Techniques

“…The predicted probability for the ℎ class given a sample vector and a weighting vector is given by (6).…”

“…The general output consists of three classes, one for each expression. In case of applying CNN the outputs were given to a fully connected hidden layer that classifies images using the softmax function earlier in (6).…”

“…Wang in 2014 for emotional recognition with an accuracy rate 62.9% over the USTC-NVIE database [5]. In 2015 a 72.4% recognition rate was achieved by Nakanishi et al [6] using a thermal dataset consists of three subjects and three facial expressions of "happy", "neutral" and "other". The introduced system uses the 2D discrete cosine transform and the nearest-neighbor criterion in the feature vector space.…”

Emotion Recognition of a Speaker Using Facial Expression Intensity of Thermal Image and Utterance Time

“…Accordingly, we adopted utterances as the key for expressing human feelings because humans tend to speak aloud when expressing their feelings. [2][3][4] In the present study, using 23 different utterance combinations based on the first and last vowels in each utterance, we investigated the performance of our previously proposed method 5 of recognizing emotions by utilizing facial expression intensity 6,7 and the utterance time.…”

“…[1][2][3][4] The timing used by a robot when attempting to recognize facial expressions is also important because the required processing can be timeconsuming. Accordingly, we adopted utterances as the key for expressing human feelings because humans tend to speak aloud when expressing their feelings.…”